AWS Adds Kimi K3.1 Pro Highspeed to Amazon Bedrock

AWS announced that Kimi K3.1 Pro Highspeed from Moonshot AI is now available on Amazon Bedrock. The model is a post-trained version of Kimi K3 featuring a 3T-class open-weight architecture and a 1-million-token context window. According to Moonshot AI, increased post-training compute targets reduced factual errors and enhanced long-horizon agentic tasks, including CAD scripting, 3D scenes, financial modeling, and game engine coding. Moonshot AI reports that output tokens generate at 2 to 3 times the speed of Kimi K3.1 Pro. The integration supports Bedrock's Flex service tier for lower inference costs, explicit prompt caching, and standard AWS enterprise data isolation.

Key takeaways

• Moonshot AI's Kimi K3.1 Pro Highspeed is available on Amazon Bedrock with a 1-million-token context window and a 3T-class open-weight architecture.
• Moonshot AI claims the model generates tokens 2 to 3 times faster than Kimi K3.1 Pro, targeted at long-horizon agentic workflows and coding tasks.
• The deployment supports configurable reasoning effort, explicit prompt caching for repeated prefixes, and discounted inference via the Flex tier.
• Inference requests run within the AWS data boundary with zero data retention enabled by default, isolated from Moonshot AI.